Locational behavior of information technology industry in the United States.

摘要

The existing body of literature on the locational behavior of firms is deficient in a number of regards. First, prior research failed to reach a consensus on which factors influence locational behavior. Second, prior research focused on traditional manufacturing, which represents a shrinking part of the economy. Third, data availability hindered the scope of research and the methodology that was employed. This paper addresses these shortcomings by using a newly available micro data set of establishment births at the county level, permitting employment of a discrete choice model.; Two estimation methods are employed. A conditional logit is used to model the influence of location-specific characteristics on firm locational behavior. Because the choice set is large (over 3,000 counties in the contiguous United States) a random draw procedure is employed to estimate a plausible behavioral model. A Poisson regression is also employed, exploiting the equivalence with the log likelihood function of the conditional logit. The random draw model is then modified to draw only counties from the same region as each chosen county. In this manner, the potential independence of irrelevant alternatives problem is mitigated. Within the Poisson model, state dummy variables are included to account for unobserved state-level characteristics.; There is a dearth of empirical research on the locational behavior of information technology industry; therefore the conditional logit and Poisson regression are employed to model the locational decisions of these I.T. firms. Information technology is divided into producers of I.T. and intensive users of I.T. The research finds that for each classification of I.T., educational attainment of the workforce has the largest impact on the probability of a county's selection by an I.T. firm. The empirical estimates further reveal that agglomeration plays a role in the locational decisions of I.T. firms. In addition, tax levels have a significant, inverse relationship on the probability of a county's selection. The results of this paper have implications for policymakers seeking to maximize economic development through public investment.